SciPy Proceedings Builder

Quickstart: Docker Container

If you already have docker engine setup, you should be able to run a version of the proceedings server by running

docker run -it -p 7001:7001 scipyproc/procbuild

Then, if you go to localhost:7001, you should see the server interface, and you can manually trigger builds by clicking the button on the right hand side.

This will occupy the terminal, when you want to shut down the server and stop running the image you will need to type ctrlC. This should shutdown the server and stop running the image.

Building the Docker image

If you want to build the Docker image from the Dockerfile contained here you should run:

docker build -t yourname/procbuild .

which will create a docker image in your local repository that you can run with:

docker run -it -p 7001:7001 yourname/procbuild

Pushing the Docker image to SciPy's DockerHub repository

Before you push the image, you need to make sure you remove the cache.

rm -rf cache

If you want to push the image to DockerHub so that it can be publicly available you would need to assign it the scipyproc/procbuild tag.

docker build -t scipyproc/procbuild .

You should also increment the version number. You do this by adding a new tag.

docker tag scipyproc/procbuild scipyproc/procbuild:x.y 

Then you should push these up to the DockerHub repository (the default repository). This should push up both the latest version and the version number.

docker push scipyproc/procbuild:latest
docker push scipyproc/procbuild:x.y

Pushing the Docker image to Heroku's repository

We run the server on in a Heroku called procbuild, found at http://procbuild.herokuapp.com.

In order to update the deployed version of the app, we need to push the image from our local repository to the heroku repository.

This requires first retagging the image, specifically we need to tag the image in a way that the heroku registry expects.

You can do this with the same tagging functionality

docker tag scipyproc/procbuild registry.heroku.com/procbuild/web

Then you need to push it to the heroku registry

docker push registry.heroku.com/procbuild/web

General notes

• Customize runserver.py to update this year's branch. For debugging, enable ALLOW_MANUAL_BUILD_TRIGGER.
• In the scipy-conference/scipy_proceedings repo: in the webhooks add a payload URL pointing to the webapp (such as http://server.com:5000/webhook). You must select only Pull Requests in the checkbox menu.
• Install dependencies: pip install -r requirements.txt
• Fetch PRs by running ./update_prs
• Launch by running runserver.py

Note: the server will run only on 3.6+

You need all the same dependencies as for building the proceedings as well.

This includes some packages on pypi which we'll install with pip. The easiest way to do this is by pulling down the latest version of the file with curl:

pip install -r <(curl https://raw.githubusercontent.com/scipy-conference/scipy_proceedings/2023/requirements.txt)

Additionally, you will need to install a version of LaTeX and some external packages. We encourage you to visit https://www.tug.org/texlive/ to see how to best install LaTeX for your system.

• IEEETran LaTeX class
  • (often packaged as texlive-publishers, or download from CTAN)
• AMSmath LaTeX classes (included in most LaTeX distributions)

If you can use apt-get, you are likely to install everything with:

apt-get install python-docutils texlive-latex-base texlive-publishers \
                texlive-latex-extra texlive-fonts-recommended \
                texlive-bibtex-extra